4 MOLECULES, VIRUSES, AND BACTERIA 



Of course, it would be more prudent to defer such an examination 

 until we were more certain of the roles of protein and nucleic acid— 

 that is, until we had such detailed knowledge that we would run little 

 risk of mistaking the principles involved. However, it seems to me that 

 we can already make intelligent guesses about the sort of way the 

 molecules function, and that it will do no harm to open the subject 

 for discussion. 



I shall assume, then, that there is some validity in the current ideas 

 about molecular biology: that almost every biochemical process is cata- 

 lyzed by a special enzyme; that all enzymes are proteins; that proteins 

 (with the exception of cross-linking by S-S bonds) are unbranched 

 polypeptides, made from a standard set of 20 amino acids; and that 

 each one has its own precisely determined amino-acid sequence and is 

 folded in a special manner which is essential for its activity. As to the 

 nucleic acids, I shall follow the current "fashionable" ideas: that they 

 constitute the most important part, if not the sole part, of the genetic 

 material; that DNA is replicated by the complementary pairing mech- 

 anism; that the genetic information lies mainly in the exact sequence 

 of bases of the nucleic acid; and that its main function is to control, in 

 some way not yet understood, the amino-acid sequence of proteins. 

 And I shall assume that the reader is tolerably familiar with all this 

 ( see the general references ) . 



Let us first examine what properties we need for natural selection 

 to operate. It is probably impossible to give an exhaustive set of ab- 

 stract postulates; rather we will generalize from the type of system 

 that nature has actually produced. 



The first requirement appears to be for specific replication— that 

 is, at some stage there must be a rather exact copying process. This 

 need not necessarily be direct: i.e., A might produce a copy B; B a copy 

 C; and then C a copy like A. But I do not think it can be entirely "arith- 

 metical." By this I mean it is not enough just to have a master copy, 

 from which are made subsidiary copies which cannot themselves be 

 copied. It is not enough to have a printing press producing newspapers. 

 There must be a mechanism by which the newspaper can be copied 

 back into type. This kind of copying, in which a copy can itself be 

 copied, I call "geometrical," and for natural selection to operate, the 

 process must be "geometrical" to some extent. The reason is obvious. 

 Natural selection has two functions. It is in part a device to enable 

 errors to be corrected ( especially errors of replication ) , and since some 

 errors are inevitable, the population must be able to eliminate them. 

 But this very process of protecting against a downward drift gives the 

 necessary mechanism for a positive evolution— i.e., not merely a stabili- 

 zation of the status quo but a progressive improvement in "fitness." In 

 order to do either of these jobs, the organism must replicate geometri- 



